Generative AI: From Jagged Frontiers to Optimised Collaborations
Knowing when to use AI is just as important as knowing how to use it.
Generative AI is the next frontier in technological innovation. However, like any frontier, it's marked by areas of both clarity and obscurity. This duality has given rise to what experts term the "jagged frontier" of AI capabilities.
The jagged frontier works as follows. On one end of the spectrum, we witness AI's remarkable prowess. Tasks that once seemed insurmountable for machines are now executed with precision and efficiency. From generating creative content to predicting complex patterns, AI has showcased capabilities that were once the exclusive domain of human cognition.
Yet, on the flip side, there are tasks where AI falters, struggling to match human intuition and adaptability. These are areas marked by nuance, context, and intricate decision-making — realms where the binary logic of machines meets its match.
The idea first emerged in the recent paper "Navigating the Jagged Technological Frontier". The work shed light on this uneven landscape, underscoring that while AI can lead to significant performance gains in suitable tasks, there are inherent risks when it's misapplied or stretched beyond its current capabilities.
For tasks inside the frontier, consultants using AI were far more productive, completing 12% more tasks 25% faster. Overall work quality was over 40% higher and the study shows that while AI benefited all users, it particularly helped lower performers whose quality increased 43%, providing more evidence that AI is the great leveller in the skills divide.
However, for a task slightly outside AI's capabilities, consultants were 19% less likely to produce correct solutions with AI. Understanding the shape of this jagged, rapidly evolving frontier is vital as businesses continue to put these tools into practice.
The key lies in discerning where AI can add value and where human expertise remains paramount. By understanding this "jagged" nature of AI's capabilities, businesses can strategically deploy AI tools, harnessing their strengths while mitigating potential pitfalls.
In order to harness the full potential of Generative AI then, the collaboration between humans and machines is proving to be a pivotal factor. But how can businesses strike the right balance, ensuring that tasks are delegated optimally between human expertise and AI capabilities?
One piece of recent research titled "Optimising Delegation between Human and AI Collaborative Agents" offers a potential solution to this conundrum. The study introduces a novel approach: a manager agent, a computational AI system designed to learn and determine the best delegation strategies between human and AI teammates, depending on the context.
This manager agent operates by modelling human and AI teammates separately, recognising their distinct capabilities and limitations. Through the power of reinforcement learning, the system discerns the optimal delegations based solely on observed outcomes, eliminating the need for intricate internal details.
While the paper focuses on the problem. It is not the only way for businesses to navigate the jagged frontier successfully. However, just like the machines do themselves, businesses must engage in a form of reinforcement learning in order to do so. By understanding that successful AI integration hinges on a symbiotic relationship between humans and machines, businesses can, through trial-and-error, see how each party (man and machine) complements each other's strengths and compensates for their weaknesses.
But of course, it is not only imperative that businesses know when to delegate to the machine, but also how to use these tools correctly when they do. Skills like Chain-of-Thought (CoT) & Chain-of-Density (CoD) prompting, for example, are proven methods of reducing hallucinations in LLMs and providing more secure and accurate results. In this sense, navigating the frontier is as much about learning how to use GenAI as much as it is understanding when you use it, else you risk sub-optimal outcomes.
The keys to the AI kingdom then are a combination of education and experimentation (which together equal augmentation). Fortunately for businesses, there is still time for both. Enterprises sit atop a world of new innovations, and it is clear that there is still ample opportunity to retain (or regain) competitive edge, for we are still at the beginning of the Generative AI technological wave.
Read more about technological waves here.
In August, a report by McKinsey showed that one-third of all organisations are already regularly using Generative AI in at least one function — and that 60% of organisations with reported AI adoption are using GenAI. What’s more, 40% of those reporting AI adoption at their organisations say their companies expect to invest more in AI overall thanks to Generative AI, with 28% saying Generative AI use is on their board’s agenda.
Just last week, the KPMG CEO Outlook for 2023 confirmed such findings, revealing that 70% of global CEOs are investing heavily in Generative AI (GenAI). This strategic bet comes despite concerns like costs (55%), regulatory challenges (50%), and cybersecurity fears (82%).
As we covered previously in “Generative AI: A Usability Challenge or a Competitive Edge?” the early stages of any technological wave are typically marked by a blend of curiosity, experimentation, and a degree of caution. By understanding the risks and how to mitigate them, organisations can judo-flip their concerns around GenAI, flipping this technological adoption from a hurdle into an hallmark of their Gen-AI enhanced workflows.
Here at ACQUAINTED, the thing that is most clear about Generative AI is that the future belongs to those who embrace the complexities and opportunities of this remarkable technology. In this sense, the "jagged frontier" of AI capabilities is not a barrier; it's a navigational challenge, and understanding where AI excels and where it seeks human guidance is the vital to unlocking its full potential. Furthermore, when you have identified when to use it, it is imperative that you understand how to use it in order to ensure safe, secure, and efficient outcomes. By strategically combining human expertise and AI capabilities, businesses can achieve remarkable things: as long as they know when and how to use this incredible technology.
In an era where GenAI's capabilities vary across a jagged frontier, here at ACQUAINTED we’ve honed our GenAI offerings to cater to your unique business needs. Whether you operate in professional services, education, or any other industry, our talks, workshops and solutions are meticulously crafted to help you navigate this jagged terrain effectively. Get in touch today.